Inhibition of BMP and of TGFβ receptors downregulates expression of XIAP and TAK1 leading to lung cancer cell death

To better understand the role of BMP inhibitors to treat lung cancers, the potency of the BMP receptor inhibitor, DMH2 [16, 31], to decrease the expression of Id1 and its effect on cell growth was examined in a panel of lung cancer cell lines of different cell types and mutations commonly found in lung cancer. The lung cancer cell lines examined consisted of a squamous carcinoma (H157), adenocarcinoma (A549), poorly differentiated carcinoma (H1299), low-grade (H727) and high-grade (U1752) neuroendocrine carcinomas. The mutations included a p53 deletion (H1299), activating K-Ras mutations (A549, H157, H727), and PNET deletion (H157). DMH2 decreased the expression of Id1 (Additional file 1: Figure S1A) and inhibited growth in all of the cell lines (Additional file 1: Figure S1B). These studies suggest that inhibition of BMP signaling may have significant anti-tumor effects across a broad range of lung carcinomas.

The effect of the BMP receptor inhibitors DMH2 and LDN-193189 (LDN) [32] to regulate Id1 expression in tumor xenografts was examined. Established xenografts from H1299 cells stably expressing an Id1 promoter regulating luciferase were treated with DMH2 or LDN and tumor luminescence was measured before and after treatment. Tumor luminescence was increased at 4 h, 24 h, and 48 h following treatment compared to baseline with the greatest effect seen at 24 h (Fig. 1a–d). In total, 13 of 17 mice treated with a BMP inhibitor showed an increase in Id1 promoter activity in the tumor (Fig. 1a–d). Immunoblot analysis demonstrated a 27 % increase in ID1 expression in xenografts treated with BMP inhibitors for 24 h as compared to controls (Fig. 1e and f). At 9 days following treatment with DMH2 there was a 103 % increase in the expression of Id1 protein compared to controls (Fig. 1g and h). We examined in vitro the dose responsive regulation of Id1 in H1299 and A549 cells following treatment with DMH2 for 24 and 48 h. These studies showed that low doses of DMH2 as low as 0.5 μM can cause an increase in the expression of Id1 followed by suppression of Id1 expression at higher concentrations (Fig. 1i and j, Additional file 2: Figure S2). In the study for the H1299 cells, we found a decrease in Id1 expression followed by an increase and complete suppression at higher doses. These studies are consistent with BMP inhibition leading to feedback activation of Id1 expression with lower concentration of BMP inhibitor.

We next examined the bioavailability of DMH2. The percent of DMH2 binding plasma proteins in human and mouse plasma is 98.6 % +/−0.7 and 98.2 % +/−1.5 respectively. DMH2 exhibited moderate to high systemic plasma clearance (68.45 mL/min/kg) with elimination half-life of 0.95 h (Additional file 2: Figure S2A). LDN is reported to have an elimination half-life of 90 min in mice [32]. The IC50 of DMH2 for the BMP type I receptor Alk2 is reported to be 42.8 nM [31]. Therefore, the free fraction of unbound DMH2 above its IC50 for Alk2 is less than 25 min (Additional file 3: Figure S3B). These pharmacokinetic studies suggest that DMH2 would provide only weak inhibition of BMP signaling in tumor xenografts. These studies suggested that DMH2 might be a weak inhibitor in vivo, which allows other signaling pathways to regulate the transcription of Id1.

To assess potential activated feedback loops following inhibition of BMP signaling, we examined the pathways reported to regulate the transcription of Id1. Src was not activated in tumors treated with BMP inhibitors for 24 h or 9 days. Although inconsistent, some of the tumors treated with BMP receptor inhibitors had an increase in the expression of phosphorylated MEK-1/2 (Additional file 3: Figure S3A and B). In vitro, DMH2 caused an increase in the expression of phosphorylated MEK-1/2 in H1299 (Additional file 3: Figure S3C) and A549 cells (Additional file 4: Figure S4D). We examined whether Egr-1, the transcriptional regulator of Id1 induced by pMEK-1/2, was regulated following BMP inhibition. We found no change in the expression of Egr-1 with the activation of MEK-1/2, suggesting MEK-1/2 did not regulate Egr-1 in the H1299 cells (Additional file 4: Figure S4C). The activation of MEK-1/2 corresponded with a decrease in the expression of Id1 (Additional file 4: Figure S4D and Additional file 1: Figure S1J) and not an increase as would be expected if it where activating the transcription of Id1. The specific MEK-1/2 inhibitor PD0325901 (PD) [33] did not regulate the promoter activity of Id1 as demonstrated in the Id-1 luciferase reporter assay in H1299 cells (Additional file 4: Figure S4E). Furthermore, PD when combined with DMH2 had no additional effects on growth inhibition of H1299 and A549 cells compared to DMH2 alone (Additional file 4: Figure S4F-G). These studies support that Scr and MEK-1/2/Egr-1 signaling pathways were not the mechanism causing the increase in Id1 expression following low levels of inhibition of BMP signaling.

Since both the BMP and TGFβ signaling cascades regulate TAK1, which has been shown to cause a feed-forward activation of BMP signaling in cartilage progenitor cells [34], we next examined the activation of TAK1 and TGFβ following the suppression of BMP signaling. Tumor xenografts treated with DMH2 for 24 h and 9 days showed activation of TAK1 and TGFβ signaling as demonstrated by an increase in phosphorylated TAK1 and Smad2 respectively (Fig. 2a and b). In vitro, DMH2 caused an increase in phosphorylation of TAK1and Smad2 at lower doses, which tapered off at higher doses (Fig. 2c and d). The decrease in activity of Smad2 corresponded with a significant decrease in the expression of Id1 in both H1299 and A549 cells (Fig. 1j and Additional file 3: Figure S3D). The unphosphorylated Smad2 was regulated in a similar fashion except that its downregulation occurred at a higher concentration than pSmad2 (Fig. 2c and d). Knockdown of the BMP type I receptors alk2 and alk3 as well as alk2, alk3, and alk6 caused an increase in expression of Smad2 and pSmad2 (Fig. 2e). DMH1, an inhibitor specific for BMP type I receptors with no activity for TGFβ receptors [31], also caused the activation of Smad2 and TAK1 (Fig. 2f). Unlike DMH2 that inhibits the TGFβ type I receptor Alk5 [31], DMH1 did not decrease the expression of Smad2 or pTAK at higher doses. These studies show that suppression of BMP signaling results in a feed-back increase in the expression and activation of Smad2 and TAK1. These studies also demonstrated a functional difference between the BMP inhibitors DMH1 and DMH2.

We previously reported that DMH2 causes a greater decrease in the expression of Id1 protein and cell growth in comparison to DMH1 [16]. To better understand differences between DMH1 and DMH2, we determined the IC50 of DMH2 for all the BMP type I receptors, Alk5, as well as BMPRII and TGFβ type II receptors and compared it to the reported literature for DMH1 and LDN (Table 1) [35]. Both DMH1 and DMH2 inhibited Alk2 and Alk3 at nanomolar concentrations. DMH2 inhibited Alk5 at an IC50 of 1690 nM, which would account for the decrease in pSmad2 expression at 2.5 μM (Fig. 2c and d). Unique to DMH2 was the inhibition of BMPRII, which does not occur with DMH1 or LDN (Table 1). DMH2 causes significantly greater suppression of Id1 promoter activity at 1.0 μM and 2.5 μM in the H1299 cells in comparison to DMH1 (Additional file 5: Figure S5). We also found that DMH2 induces significantly greater growth suppression and cell death of H1299 cells compared to DMH1 and LDN (Additional files: Figure S6 and Figure S5). These data suggest that the inhibition of both the BMP and TGFβ signaling cascades may enhance the downregulation of Id1 leading to greater growth inhibition and cell death.

We next examined whether TAK1 causes a feed-forward activation of BMP signaling in lung cancer cells and whether both the BMP and TGFβ signaling pathways activated TAK1. Transient expression of constitutively active BMP and TGFβ type I receptors caused the activation of TAK1 and Id1 expression in the H1299 cells (Fig. 3a and b), confirming their role in activating TAK1 in our lung cancer cells. The irreversible TAK1 inhibitor, 5Z-7-oxozeaenol (5Z) [36], causes a dose-related decrease in the expression of pSmad 1/5, Id1, and pTAK1 after 24 h (Fig. 3c). After 24 h, 2 μM of 5Z also suppressed Id1 promoter activity (Fig. 3d). Interestingly, after 48 h 5Z had the opposite effect, now increasing the expression of Id1, pSmad 1/5, and pTAK1, and no longer suppressed the Id1 promoter (Fig. 3e and f), demonstrating that 5Z induces feedback activation of its intended target. Furthermore, the combination of 5Z and DMH2 also did not enhance growth suppression (Fig. 3f–h). These data suggest that activated TAK1 stimulates Smad 1/5-Id1 signaling through a feed-forward mechanism in lung cancer cells.

We next examined TGFβ regulation of TAK1 and Id1 using the highly specific TGFβ antagonist SB-505124 (SB) [37] alone and in H1299 cells in which BMP signaling was suppressed with low dose of DMH2. SB alone caused a dose-related increase in the expression of Id1, suggesting TGFβ suppressed Id1 expression in H1299 cells (Fig. 4a). When BMP signaling was suppressed, SB had the opposite effect, now causing a dose-related decrease in the expression of Id1 (Fig. 4a). TGFβ signaling is reported to increase the expression ATF3, which is a co-repressor required for Smad3 to suppress the Id1 promoter [21]. In the absence of ATF3, Smad3 can activate the transcription of Id1. There were no significant change in the expression of ATF3 following suppression of BMP and TGFβ signaling, suggesting that ATF3 was not involved in the change of Id1 expression (Fig. 4a). SB alone caused an increase in the Id1 promoter activity in H1299 cells (Fig. 4b). When BMP signaling was suppressed with a low dose of DMH2, SB again had the opposite effect, now causing a dose-related decrease in Id1 promoter activity (Fig. 4b). We next asked whether inhibiting the feed-back activation of TGFβ signaling would enhance DMH2 downregulation of Id proteins and pTAK1. DMH2 caused a greater downregulation of the expression of Id1 and Id3 when used in combination with SB compared to DMH2 alone (Fig. 4c). In addition, there was a greater decrease in the expression of pTAK1 when DMH2 and SB were used in combination (Fig. 4c).

To further examine the role of TGFβ signaling causing a feedback activation of the BMP-TAK1-Id1 signaling cascade, H1299 cells were transiently transfected with vector control or constitutively active Alk5 (caAlk5) and treated with 0.1 and 0.5 μM of DMH2 for 48 h. Low doses of DMH2 were used so as not to inhibit TGFβ signaling. When BMP signaling was inhibited by DMH2, caAlk5 enhanced the activation of TAK1 and Id1 expression compared to cells treated with DMH2 alone (Fig. 4d). The increase in expression of Id1 and pTAK1 induced by caAlk5 together with 0.5 μM of DMH2 was antagonized by a 24-h treatment with 5Z (Fig. 4d). These studies support our observation that when BMP signaling is attenuated, activation of TGFβ signaling increases Id1 expression by activating TAK1.

If inhibiting both BMP and TGFβ signaling enhances the downregulation of Id1, we would expect to have greater growth suppression when both were used together. We examined whether suppressing both BMP and TGFβ signaling enhanced growth suppression by treating cells with a BMP receptor inhibitor alone or in combination with SB. DMH2 alone caused significant growth suppression of the H1299 (Additional file 6: Figure S6A). The combination of DMH2 and SB caused a significantly greater decrease in cell number compared to either drug alone (Additional file 6: Figure S6A). SB alone had little effect on the growth of the H1299 cells (Additional file 6: Figure S6A). In the A549 cells, both DMH2 and SB-505124 alone caused growth suppression (Additional file 6: Figure S6B). A549 cells treated with both 2.5 μM DMH2 and 2.0 μM SB had significantly fewer cells than either compound alone (Additional file 6: Figure S6B). 2.5 μM of DMH1 combined with 1.0 μM or 2.0 μM SB-505124 caused significantly greater growth suppression than either compound alone (Additional file 6: Figure S6D).

We next examined if the regulation of TAK1 was a mechanism by which BMP inhibitors induced death of lung cancer cells. DMH2 induced expression of activated caspase-3 at 48 h (Fig. 5a and b) and by 72 h cells showed significant shrinkage and chromatin condensation (Fig. 5c). These studies support our observation that DMH2 induces apoptotic cell death. DMH2 induced significant cells death of H1299 and A549 cells after 3 days (Fig. 5d and e). The addition of SB together with DMH2 did not cause further cell death (Fig. 5d and e). Examining the mechanisms by which DMH2 induces cell death, we found that DMH2 caused a very significant decrease in the expression of Smad2 in both H1299 and A549 cells (Fig. 5f and g). This was associated with a decrease in expression of TAK1 and the phosphorylation of its downstream target the NF-kappa B subunit p65 [28] (Fig. 5f and g). Knockdown of TAK1 with siRNA in the H1299 cells caused cell death suggesting its downregulation is a mechanism by which DMH2 induces cell death (Fig. 5h and i). DMH1 (2.5 μM) alone or when combined with SB caused little cell death in comparison to DMH2 (Fig. 5j). DMH1 alone or in combination with SB also did not activate caspase-3 or downregulate Smad2, pTAK1, or p-p65 after 3 days (Fig. 5k and l). LDN is an inhibitor of both the BMP and TGFβ receptors. LDN also caused little cell death and downregulation of Smad2 and TAK1 in comparison to DMH2 after 3 days (Fig. 5m and n). These studies supports that DMH2 downregulation of TAK1 is involved in induction of cell death but that additional mechanism(s) may also be involved in addition to inhibition of BMP type I receptors and TGFβ signaling.

During embryonic development the BMP signaling cascade mediates the activation of TAK1 by stabilizing the expression of XIAP through Smad1/5 independent mechanism [38]. DMH2 caused a significant decrease in the expression of XIAP in both the H1299 and A549 cells after 3 days (Fig. 6a and b). LDN and DMH1 alone or with SB-505124 did not downregulate the expression of XIAP after 3 days (Fig. 6c and d). Knockdown of XIAP with siRNA in the H1299 cells induced cell death (Fig. 6e) and caused a decrease in expression of pTAK1 (Fig. 6f). Since BMP type I and type II regulate the ubiquitination of XIAP [26], mutant XIAP that has had its lysine ubiquitination sites mutated [39] was transiently transfected into the H1299 cells. H1299 cells expressing mutant XIAP had a higher expression of pTAK1 compared to vector control cells (Fig. 6g). DMH2 decreased the expression of Id1 and wild type XIAP but not the mutant XIAP (Fig. 6h). Mutant XIAP and vector control cDNA were stably transfected into the H1299 cells. H1299 cells expressing mutant XIAP were resistant to cell death induced by DMH2 (Fig. 6i). In cells stably expressing mutant XIAP, DMH2 also decreased the expression of wild type XIAP but not mutant XIAP (Fig. 6j). These studies show that XIAP is an upstream regulator of TAK1 in lung cancer cells and that the downregulation of XIAP is required for DMH2 induced cell death.

To further examine the role of the TGFβ signaling cascade in regulating cell death, XIAP was knocked-down in H1299 cells then cells were treated with the inhibitor of the BMP type I receptors (DMH1) and with the TGFβ antagonist LY2109761 (LY) [40]. The greatest cell death and decrease in pTAK1 expression occurred in XIAP knockdown cells treated with a combination of DMH1 and LY (Fig. 7a and b). These data further support that the downregulation of XIAP and inhibition of TGFβ signaling are mechanisms by which DMH2 inhibits TAK1 and induces cell death.

Constitutively active alk3 and alk6 constructs in mammalian vectors were gifts from Joan Massague (New York, New York). Constitutively active alk5 was a gift from Fang Liu (Rutgers-New Jersey Medical School). The Id1/luciferase promoter was a gift from Dr. Desprez (California Pacific Medical Center). Plasmids pc DNA3-XIAP-Myc and pcDNA3-XIAP-Myc K322/D28 were gifts from Guy Salvesen (Addgene plasmid # 11833 and plasmid # 11834).

The A549 and H1229 lung cancer cell lines were cultured in Dulbecco’s Modified Eagle’s medium (DMEM, Sigma Aldrich, St Louis, MO, USA) with 5 % fetal bovine serum (FBS), 1 % antibiotic/antimycotic, and 1 % L-glutamine [52]. The lung cancer cell lines H157, H727, U1752, and H358, and H865 were cultured in 90 % RPMI and 10 % FCS. The cell lines were obtained from ATCC and from Malcolm Brock, Johns Hopkins University.

Select siRNA were used to target the type I BMP receptors alk2, alk3, and alk6 (Life Technologies). The ID numbers of the siRNA are: alk2 (s974), alk3 (s281), alk6 (s2042). The knockdown of the BMP receptors with these siRNAs have been previously validated on the H1299 cells and confirmed on the blots used in this study [16]. Validated siRNA XIAP (S1456), siRNA TAK1, and siRNA BMPRII were purchased from Life Technologies. Silencer Select Negative Control siRNA (4390843) was used to confirm specificity of each targeted knockdown.

Cells were transfected with siRNA using a Nucleofector II (Amaxa Biosystems, Gaitherburg, MD) using the manufacture’s Nucleofector kit T. Studies using cell death assay cells were transfected with Lipofectamine 3000. A total of 30nM of siRNA was used for alk2 and alk3 [16] and 20 nM for alk6 and 30nM for XIAP. Plasmid cDNA was transfected at 2.5ug.

Total cellular protein was prepared using RIPA buffer containing a protease inhibitor cocktail and protein concentration was measured using the BCA assay as described [4]. In brief, protein was analyzed by SDS-PAGE, transferred to nitrocellulose (Schleicher and Schuell, Keene, NH). After blocking, the blots were incubated overnight at 4 °C with the appropriate primary antibody in Tris-buffered saline with 1 % Tween (TBST) and 5 % BSA. Secondary antibodies were applied for 1 h at room temperature. Specific proteins were detected using the enhanced chemiluminescence system (Amersham, Arlington Heights, IL). The primary antibodies that were used were rabbit monoclonal anti-pSmad 1/5, rabbit monoclonal anti-pSmad2, rabbit monoclonal anti-Smad2, rabbit monoclonal anti-pTAK1, rabbit monoclonal XIAP, rabbit anti-monoclonal p-p65, rabbit monoclonal anti-activated caspase-3, rabbit polyclonal anti-BMPRII, and rabbit monoclonal anti-EGR-1 (Cell signaling Technology, Danvers MA), Rabbit monoclonal anti-TAK1 (Invitrogen, Grand Island NY), rabbit anti-actin, an affinity isolated antigen specific antibody (Sigma, Saint Louis, MO), rabbit monoclonal anti-Id1 and rabbit monoclonal anti-Id3 (Calbioreagents, San Mateo, CA), rabbit polyclonal anti-GAPDH (Sigma, St. Louis, MO) and rabbit polyclonal pMEK-1/2 (Cell Signaling, Danvers MA).

H1299 cells were stably transfected with the Id1-promoter luciferase reporter. Treated cells were lysed with 1X luciferase lysis buffer (Promega). Samples were added to luciferase assay substrate (Promega) and luminsescence measured by the TD-20/20 Luminometer (Turner Designs/Turner BioSystems, Sunnyvale, CA). Control samples included luciferase assay substrate alone and luciferase assay substrate plus 1X reporter lysis buffer.

H1299 cells were stably transfected with the Id1 promoter-luciferase reporter. Five million cells were injected subcutaneously into the flank of NCI female nude mice. After approximately 12 days the mice were anesthetized and the luciferase activity measured. Baseline luciferase activity was obtained just prior to the injection of inhibitors. Mice were injected IP with 150 mg/kg luciferin-D and after a 15-min uptake period tumors were imaged with IVIS Spectrum and analyzed using Living Image software. Mice were injected IP with DMSO, 3 mg/kg DMH2, or 3 mg/kg LDN and after 4 h luciferase activity determined. Two days later, baseline luciferase activity again determined and mice were injected IP with DMSO, 3 mg/kg DMH2, or 3 mg/kg LDN every 8 h for three doses and luciferase activity was measured 4 h after the last dose. In separate animal experiment, mice with established tumors were treated with DMSO or 3 mg/kg DMH2 twice daily for 48 h and luciferase activity compared to baseline. The tumors were harvested on ice, homogenized, and placed in lysis buffer. In a separate experiment, mice with established tumor xenografts were injected IP with DMSO, 3 mg/kg DMH2, or 9 mg/kg DMH2 every 12 h for 9 days and tumors were then harvested.

Cells were plated into 6 well plates at 10⁵ cells per well and treated with 1 μM DMSO or antagonist for 7 days. The cells were detached with trypsin, stained with trypan blue, and the number of live cells counted using a hemocytometer.

Cells were plated in 6 well plates with 10⁶ cells per well. Three days following treatment with inhibitor(s) and/or transfection the adherent and floating cells were harvested and incubated with 0.1 mg/ml of ethidium bromide. Cells that die by necrosis or apoptosis lose membrane integrity and take up ethidium bromide [53]. Immediately after staining approximately 100 cells were counted and the percentage of cells that took up ethidium bromide was determined.

The IC₅₀ of DMH2 for alk2, alk3, alk6, alk5, BMPRII, and TGFβ were performed at Reaction Biology Corporation (Malvern, PA). This was a 10-point assay starting from 100 μM to 100 nM performed in duplicate. The ATP concentration was 10 Micromolar.

Human and mouse protein plasma binding of DMH2 was performed using equilibrium dialysis (Sai Life Sciences Limited, Pune India).

The pharmacokinetics of DMH2 was examined in maie BALB/c mice following intravenous and oral administration (Sai Life Sciences Limited, Pune India). Three mice at each time point were dosed with a 2 mg/kg tail vein injection and 10 mg/kg p.o. Blood samples were taken 0.08, 0.25, 0.5, 1, 2, 4, 8, 12, and 24 h (i.v.) and 0.25, 0.5, 1, 2, 4, 6, 8, 12, and 24 h (p.o.). Plasma half-life, clearance, and volume of distribution were then determined.

The mean of the control group was compared to the mean of each treated group using a paired student t-test assuming unequal variances. Differences with p values <0 .05 were considered statistically significant.